VARIABILITY IN LIME-MUD FEED RATE INDUCED IN THE LIME … · in the lime kiln [3]. Furthermore,...

VARIABILITY IN LIME-MUD FEED RATE INDUCED IN THE

Downloaded from: https://research.chalmers.se, 2020-01-19 18:38 UTC

Citation for the original published paper (version of record):Dernegård, H., Berg, E., Brelid, H. et al (2017)VARIABILITY IN LIME-MUD FEED RATE INDUCED IN THE LIME-MUD FILTER:

N.B. When citing this work, cite the original published paper.

research.chalmers.se offers the possibility of retrieving research publications produced at Chalmers University of Technology.It covers all kind of research output: articles, dissertations, conference papers, reports etc. since 2004.research.chalmers.se is administrated and maintained by Chalmers Library

(article starts on next page)

ABST

RACT

19J-FOR Journal of Science & Technology for Forest Products and Processes: VOL.6, NO.3

TRADITIONAL AREA CONTRIBUTIONS

This study shows that despite keeping both lime-mud slurry flow and density constant, lime-mud moisture content and feed rate from the filter to the lime kiln can vary significantly. These variations, which were previously not detected in the mill, were revealed after installation of a band scale and a moisture detector. It was found that cleaning the doctor blade caused a periodic disturbance in the lime-mud feed rate to the kiln. The variations in moisture content, however, were too minor to disrupt the stability of a kiln equipped with a flash dryer. This work resulted in a new and better way of continuously estimating available CaO and reburned lime production rate.

HENRIC DERNEGÅRD*, HARALD BRELID, HANS THELIANDER, ERIK BERG

VARIABILITY IN LIME-MUD FEED RATE INDUCED IN THE LIME-MUD FILTER: CONSEQUENCES AND ACTIONS

The present investigation was carried out at Södra Cell Mönsterås, Sweden. The mill produces 750,000 ADt of fully bleached Kraft pulp per year. The mill produces softwood and hardwood pulp in cam-paigns that are around 25 and 10 days long, yielding 550,000 ADt softwood pulp and 200,000 ADt hardwood pulp.

The mill has two lime kilns equipped with lime-mud flash dryers. Both kilns are fired with approximately 70% pulverized bark and 30% tall-oil pitch. On rare occa-sions, the kilns are fired with 100% fuel oil. Kiln No. 1 is 113 metres long, has an inner diameter of 3.3 metres, and produces 450 tons/day of reburned lime. Kiln No. 2 is 73.5 metres long, has an inner diameter of 2.8 metres, and produces 250 tons/day. Kiln 1 is equipped with a labyrinth cooler, and kiln 2 has satellite coolers. The calcu-lated adiabatic flame temperature is about 1800°C for dry bark fuel, and the flue-gas temperature is about 700°C at the lime-mud feed end.

Table 1 gives a mill balance for chem-ical recovery. The amount of lime needed for the desired alkali production is around 0.28 ton of reburned lime per ADt. How-ever, the reburned lime production figure from the mill data system is 0.32 ton of

reburned lime per ADt. A dump test (all reburned lime was taken out and weighed during 30 minutes) was performed to vali-date the signal from the mill data system. The result was that the mill data system showed an overestimation of around 4.3% (0.306 ton/ADt instead of 0.32). Accord-ingly, there seems to have been an error in how the reburned lime production was calculated.

The production rate of available CaO is important for the performance of the white liquor preparation plant. An overcharge of available CaO to the slaker results in overliming, which is known to impair lime-mud filterability and cause se-vere white liquor production disturbances

INTRODUCTION and even pulp production shutdowns [1]. The production rate is required to regulate fuel demand because steady state is first reached around 90 minutes after a produc-tion change. Fuel demand after a produc-tion change must be estimated using the production rate. This is even more impor-tant when using a lime-kiln control strat-egy. Lime calcination requires 3 MJ/kg of formed CaO at 900°C [2] plus additional heat for bringing the system up to tempera-ture. A heat and mass balance at Södra Cell Mönsterås indicated that drying of lime mud requires around 1.5 MJ/kg. The heat losses through the kiln shell were around 1.1 MJ/kg, and the losses with the hot flue gas to the electrostatic precipitator were

*Contact: [email protected]

HENRIC DERNEGÅRDSödra Cell Mönsterås, Box 501, 38325 Mönsterås, Sweden

HARALD BRELIDSödra Innovation, SE-43286Väröbacka, Sweden

HANS THELIANDERChalmers University of Technology, SE-412 96 Gothenburg, Sweden

ERIK BERGOptimation AB, Box 27 941 21 Piteå, Sweden

20 J-FOR Journal of Science & Technology for Forest Products and Processes: VOL.6, NO.3

around 1.4 MJ/kg of reburned lime prod-uct (total 7 MJ/kg reburned lime). It is of the utmost importance to keep the lime-kiln temperature profile stable over time because temperature fluctuations are known to cause recarbonation of the rings in the lime kiln [3]. Furthermore, perfor-mance variations in the lime-mud filter are known to influence reburned lime quality [4,5].

This paper proposes a new procedure for calculating reburned lime production. This calculation requires measurements of mass flow, moisture content, and carbon content of the lime mud and the reburned lime. Södra Cell in Mönsterås invested in a band scale, an NIR detector, and a car-bon content analyzer to set up this new calculation procedure and to achieve bet-ter control of some of the process param-eters affecting reburned lime production. After the band scale and the NIR mois-ture detector were installed, fluctuations in lime-mud production and moisture content were noted, and a more detailed

study of the reasons for and consequences of these fluctuations was performed.

CALCULATION OF REBURNED LIME PRODUCTION Old Procedure The conventional way to calculate lime production uses the volumetric flow rate q and the density of the lime-mud slur-ry that is pumped to the lime-mud filter. The volumetric ratio, y, is calculated us-ing the linear equation, y = kx +m, where the slope coefficient, k, is inside brackets in Eq. (1). The density of the solid mate-rial in the lime mud is denoted as ρmud, and the filtrate density is denoted as ρfiltrate. The constant m is obtained by setting the slurry density ρslurry equal to ρmud with y equal to 100%:

(1)

Equation (2) shows a common calculation of lime production. The ex-

pression inside brackets gives the volume fraction of lime mud in the slurry. The factors 24 and 3.6 are used to convert kg/s to tons/day:

(2)

The lime mud-to-reburned lime con-version factor is calculated by adding the parts of the reburned lime together as in Eq. (3):

(3)

If the conversion factor is set to 0.6 (corresponding to around 93% CaCO3 in lime mud and 3% residual CaCO3 in reburned lime), ρmud is set to 2 717 kg/m3, and ρfiltrate is set to 1000 kg/m3, Eq. (2) can be simplified to Eq (4), which is

TABLE 1 Mill balance giving an estimate of effective alkali and CaO demand calculated from pulp production. ObjectPulp productionEffective alkali to digester (as NaOH)Effective alkali to white liquor oxidation Sulphidity (on active alkali)Available CaOCaO in lime mud Pulp yield, fully bleachedPulp productionWood demandEffective alkali to digester (as NaOH)Total EA demandNaOH demandNaOH demandCa(OH)2 demandCaO demandCaO productionReburned lime productionReburned lime for overlimingSpec. CaO demandSpec. reburned lime demandSpec. reburned lime demand (with 1.5% CaO in mud)

Number2,150

204038871.545

1,9354,30086094672418.19.19.150758515

0.240.270.28

UnitsADt/day

% on dry woodkg/ADt

%%%%

BDt/daytons dry wood/day

tons/daytons/daytons/dayMmol/dayMmol/dayMmol/day

tons CaO/daytons lime/daytons lime/daytons lime/ADttons lime/ADttons lime/ADt

CalculationIndata, weighedIndata, fl ow meterIndata, fl ow meterIndata, titrationIndata, laboratory analysisIndata, laboratory analysisIndata, calculation2150 × 0.91935/0.454300 × 0.2860 + 40 × 2150/10002 × (1-0.38) × 946/(2-0.38)724 / 4018.1 / 218.1 / 2 9.1 × 56507/0.87585/0.57* × 0.015507/2150585/21500.27 + 15/2150

*0.57 is a factor for converting lime mud to reburned lime, assuming that 50% of the dead load in the lime mud remains in the reburned lime. See Table 2 for more details on this calculation.



the calculation used in the mill data system at Södra Cell Mönsterås:

(4)

New procedure Analyses show that the conversion factor (g reburned lime/g lime mud) in Eq. (3) can vary between 0.55 and 0.59, with an average of 0.57 during normal conditions (a simple improvement would be to sub-stitute 0.6 for 0.57). Analyses of 11 dif-ferent samples of lime mud after the filter gave an average solid material density of 2600 kg/m3 with a standard deviation of 18 kg/m3. This is taken into account when calculating the reburned lime production (tons/day) in Eq. (5) as the sum of the cal-culated CaO, the remaining dead load, and the residual CaCO3 in the reburned lime:

(5)

CaO production in Eq. (6) is calcu-lated from the dry lime-mud feed rate, m (tons/hour), multiplied by a factor includ-ing calcination degree, cd (explained in Eq. (7)), and the CaCO3 in the mud plus the CaO already present (from overliming) in the mud. The factor 24/1000 is used to convert kg/hour to tons/day. The CaCO3 in the mud is calculated from the carbon content in the mud under the assumption that all carbon is present as calcium car-bonate. The factor 56/100 is the molar weight of CaO divided by that of CaCO3. The dry lime-mud feed rate is calculated using the moist lime-mud feed rate and the moisture content:

(6)

Two equations are required to derive the calcination degree. First, it is assumed that dry lime mud consists of CaCO3, CaO, and dead load (the balance). The second equation results from assuming that the residual CaCO3 in the reburned

lime can be calculated by subtracting the equivalent CaCO3 that was converted us-ing the calcination degree from the origi-nal CaCO3 in the lime mud. This gives an expression for residual CaCO3 in reburned lime as a function of calcination degree, CaCO3 in lime mud (actual carbon in lime mud), and CaO in lime mud. Table 2 pres-ents the concept of the derivation.

After some rearrangements, an ex-pression is obtained for the calcination degree as a function of residual CaCO3 in reburned lime, CaCO3 in lime mud, and CaO in lime mud, as shown in Eq. (7). DL represents the assumed dead load in the lime mud that remains in the reburned lime.

An analysis of the relative influence of the different components presented in Fig. 1 shows that the CaO in mud has practically no influence and can be ne-glected. Equation (7) can then be simpli-fied to Eq. (8).

The dead load production (tons/day) used in Eq. (5) is calculated in Eq. (9) us-ing the dead load in the lime mud and the

TABLE 2Derivation of an expression for residual CaCO3 depending on calcination degree. This expression can then be rearranged and used to compute the calcination degree, as shown in Eq. (7).

AssumptionsCalcination degreeNPE remaining in limeIndataResidual CaCO3 in limeCaO in lime mudC in lime mudLime mud balanceCaCO3CaO in lime mudDead load (balance)TotalReburned lime balanceCaO from calcinationR-CaCO3CaO from CaO in lime mudDead loadTotalCalculation resultCalc. R-CaCO3 in lime

98.1 50

3.01.5

10.8

9001585

1000

494.417.115.042.5

569.0

3.0

%%

%%%

g/kgg/kgg/kgg/kg

g/kgg/kgg/kgg/kgg/kg

%

Initially assumed Empirically determined, explained further on

Not used in starting balanceLaboratory analysisLaboratory analysis

10.8×10/12×1001.5×101000-900-15

900×56/100×98.1/100 900-900×98.1/100 1.5×1085×50/100 494.4+17.1+15.0+42.5

17.1/569.0×100

(7)


fraction of this dead load remaining in the reburned lime:

(9)

The dead load in the lime mud is calculated by subtracting the CaO and CaCO3 in lime mud from the total weight of dry lime mud. The residual CaCO3 in reburned lime is calculated in Eq. (10) us-ing the calcination degree:

(10)

Table 3 shows an example of the re-sults obtained by these calculations.

VALIDATION OF EXPERIMENTAL SET-UP AND CALCULATION

The band scale was a Kn4-200 instrument from SEG, Sweden, which was installed on the lime-mud conveyor belt after the filter. The band scale was calibrated with known weights on a moving belt with a deviation of 0.13%. An NIR detector from NDC Technologies, USA, was installed above

Fig. 1 - Graphical depiction of the impact of R-CaCO3 (A) and CaO in mud (B) on calcination degree as calculated using Eq. (7).

(8)

TABLE 3 New calculation of reburned lime production and available CaO. All carbon in lime mud is assumed to be present as CaCO3.

Notation

DLqρslurry

m

cd

Category IndataDead load in mud remaining in limeLime mud feed rate to fi lterSlurry density to fi lterMoisture contentWet mud feed rate to lime kilnCarbon content in dry lime mudResidual CaCO3 in reburned limeCaO in lime mud

Initial calculationsCarbonate contentLime-mud feed rate

Lime-mud mass balanceCaO in lime mudCaCO3 in lime mudDead load in lime mudSum

Reburned lime calculation Calcination degree

Reburned lime mass balanceDead load in reburned limeCaO in reburned limeCaCO3 in reburned lime

Mud-lime conversion factor

Results CaO productionDead load prod. in reburned limeR-CaCO3 prod. in reburned limeReburned lime prod. (new calc.)Reburned lime prod. (old calc.)Calc. R-CaCO3 in reburned limeAvailable CaO in reburned lime

Units

%litres/skg/litre

%tons/hour

%%%

mol/kg dry mudt/h (dry weight)

g/kg dry mudg/kg dry mudg/kg dry mudg/kg dry mud

%

g/kg dry mudg/kg dry mudg/kg dry mud

g lime/kg dry mud

tons/dtons/dtons/dtons/dtons/d

%%

Calculation

AssumedFlow meterDensity meterNIR or laboratoryBand scaleLaboratory analysisLaboratory analysisLaboratory analysis

10.8×10/1221.4× (100-24.0)/100

1.5/100×10009.0×100 (from molar conc. of C)1000-900-15

See Eq. (7)

85×50*/100900×98.1/100/100×56+15 (1-98.1/100)×900

43+509+17

16.3×509.4/1000×24**85/1000×16.3×50*/100×24** 900× (1-98.1/100) ×16.3/1000 ×24**198.9+16.6+6.717.1/569.0x100 See Eq. (4) 17.1/569.0x100 509/569x100

*Fraction of dead load entering in the lime mud (50%) that remains in the reburned lime. See the section below for further details.**The factor of 24 converts tons/hour to tons/day.

Value

50*11.11.2424.021.410.83.01.5

9.016.3

1590085

1000

98.1

42.5509.4 17.1

569.0

198.916.66.7

222.2 2183.090



the lime-mud conveyor belt after the filter. The NIR measurement was validated by capturing the mud by skimming the top few centimetres of mud traversing the conveyor belt using a 2-litre wide-mouth bottle. The collected sample was assumed to have the same mud particle distribution as that detected with the NIR measure-ment. The results obtained showed good agreement with observations and were in accordance with previous findings from another installation [6]. An ELTRA CS-800 analyzer was used to analyze the car-bon content. In this analyzer, the carbon content is determined by combustion of the solid material in an induction furnace under a pure oxygen atmosphere, oxidiz-ing carbon to CO2, which is detected with infrared cells. Initial concerns were that the carbon analysis would measure car-bon that was not present as CaCO3 and therefore would give a residual carbon-ate content that was too high. The tradi-tional method for determining CaCO3 in reburned lime at the mill includes acidifi-cation of a sample of reburned lime and measurement of the volume of released CO2 (cf. ISO 10693:2014). This is referred to as the calcimeter method. As seen in Fig. 2, the elemental analysis in fact gave a lower CaCO3 content than the calcimeter, implying a low risk of measurement arte-facts. This is a positive discovery because

the reliability of carbon detection is excel-lent and the test can be done with 0.5% accuracy for carbon content.

To validate the new calculations, in-dustrial green and white liquor samples were analyzed by ABC titration accord-ing to SCAN-N 30:85. The samples were collected and analyzed using a robot, and sampling was done approximately every 20 minutes. Na2SO4 and Na2S2O3 were analyzed by an external laboratory. The

liquor composition was used to calculate the ionic strength and the equilibrium con-stant of the causticizing reaction [7]. The equilibrium constant was used to calculate the theoretical charge of reburned lime required to reach the measured effective alkali concentration in the white liquor. The theoretical charge was compared with the results of the new production calcula-tion in Eq. (5) and the old calculation in Eq. (4).

Fig. 2 - Comparison of elemental carbon analysis and calcimeter results: (A) results from the two methods are compared in the laboratory; (B) results from the mill data system (after 2015-12-12, the carbon analyzer was used).

TABLE 4

Comparison of old and new procedures for calculating reburned lime production with a theoretical calculation of the reburned lime charge required to reach the measured effective alkali concentration in the white liquor. Na2CO3 was multiplied by 1.325 and Na2S was multiplied by 0.975 to obtain the amount as a substance. The green liquor density was measured as 1.15 kg/l at 90°C and 1.2 at 20°C. The green liquor composition was analyzed with ABC titration, and samples were collected approximately every 20 minutes. The specifi c periods were chosen to be stable and representative, with no purchased lime added.

StartFinishGreen liquorNaOH Na2CO3Na2SNa2SO4Na2S2O3GL feedIonic strengthEquilibrium constant

White liquorEffective alkaliThere of NaOH

Reburned lime and lime mudCaO in lime mudAvailable CaOProductionOld calculationNew calculationTheoretical charge

g NaOH/lg NaOH/l g NaOH/l g/kgg/kgl/smol/kg-

g/lg/l

%%

t /dt /dt /d

2017-01-152017-01-22

7.5 (5.9)105 (81.3)44.2 (34.3)

8.95.998

4.6829.43

116 (89.9)94 (72.8)

187

677573589

Comments

ABC titration as NaOH (Na2O)ABC titration as NaOH (Na2O)ABC titration as NaOH (Na2O)Only one sampleOnly one sampleFlow meter0.5 x ∑m x z2 of all ionsSee [7]

ABC titration as NaOH (Na2O)ABC titration as NaOH (Na2O)

Laboratory analysisLaboratory analysis

From mill data systemAssuming 50% of dead load remaining Assuming overliming to reach CaO in lime mud

2016-12-252017-01-01

8.0 (6.2)108 (84.0)36.6 (28.4)

8.95.91014.6329.95

116 (88)98 (76)

0.7587

757654602


The results in Table 4 show that for both time periods, the old calculation pro-cedure overestimated lime production and that the new calculation procedure gave results closer to the theoretical charge.

In the calculations in Table 3, the fraction of dead load remaining, DL, in Eqs. (7) and (9) was set to 50%. To jus-tify the choice of 50%, calculated val-ues of available CaO at different frac-tions of remaining dead load were compared with results from laboratory

analyses. Figure 3 shows that if this frac-tion is set to 100%, then the available CaO is underestimated, whereas a satisfactory fit is found when the fraction is set to 50%. A tentative explanation for dead load not being conserved in calcination could be that some metal hydroxides dissociate into metal oxides and water.

INDUSTRIAL OBSERVATIONS

Oscillations in Lime-Mud Moisture ContentWhen using the NIR signal to monitor the moisture content of lime mud after the lime-mud filter, variations in mois-ture content with a periodicity of one hour and an amplitude of 0.5–1 percent-age points were found, as shown in Fig. 4. The fluctuations coincided with the move-ment of the water jet for precoat removal, which traversed from the right-hand side of the drum towards the left-hand side at a speed of 82 mm/min while remov-ing old precoat with a water pressure of 6 MPa. The length of the filter was 9000 mm, and two water jets traversed on the same rack, meaning that one jet started at 0 mm and the second started at 4500 mm. When the whole length of the drum had been covered, the water jet traversed back to the right-hand side of the drum at a speed of 500 mm/min with a spraying pressure of only 0.6 MPa. The initial sus-picion was that the variations were due to measurement artefacts in which the lime-mud surface was perhaps moister than the bulk flow. To investigate whether these observed variations were real, mud sam-ples were taken every 10 minutes for two hours. The samples were taken by holding a 2-litre wide-mouth bottle and catching the mud that fell at various points on the drum. Care was taken to ensure that the

Fig. 3 - Comparison between laboratory analyses and predicted available CaO concentrations calculated with different lime-mud dead load fractions remaining in reburned lime.

Fig. 4 - Validation of moisture content fluctuations. Samples were taken on 2016-03-03 (4a) from 09:40 to 11:30 and on 2016-05-03 (4b) from 11:40 to 13:30 from across the entire filter.



full length of the drum was sampled dur-ing the same length of time while filling the bottle to the same degree every time to capture a representative sample. Figure 4 shows the results, which indicate that the fluctuations in moisture content are indeed real and follow the movement of the water jet.

Effects of small variations in mois-ture contentFigure 5 shows a time period during which lime production was stable and the fuel supply consisted of only fuel oil and

remained constant. Data from this pe-riod show that the feed-end temperature dropped around 5°C as lime-mud mois-ture content increased from around 25% to around 28% (1.7°C per percentage point).

Larger variations in dry solids feed rateAfter the band scale had been installed, it was noted that periods around four min-utes long of increased lime-mud feed rate occurred every eight hours. An investiga-tion of the control system showed that

the only control function with an eight-hour periodicity was a water valve that opened to wash the doctor blade (Fig. 6a). When the valve opened, the lime mud was washed away from the blade and ended up together with the wash water in the tray, thus increasing the level in the tray and the lime-mud uptake of the filter. These short periods of increased lime-mud feed were not captured using the old calculation for reburned lime production (Eq. (4)) be-cause it used the density and volumetric flow of the slurry before the filter.

During the increase in lime-mud feed rate that started at 02:40 (Fig. 6b), an ad-ditional 118 kg of dry lime mud was intro-duced into the lime kiln. The consequence of this increased lime feed was a sudden temperature drop from around 700°C to 670°C in the feed end of the kiln, as shown in Fig. 6b. The reason for the tem-perature drop was most likely an increase in the amount of water in the lime mud. The negative spike in feed-end tempera-ture could potentially cause problems with recarbonation due to a change in the tem-perature profile in the lime kiln.

Comparison with dynamic simulationsMill data were compared with findings from dynamic simulations (Language: Mod-elica, simulation tool: Dymola from Das-sault systems) performed by Optimation AB, Sweden. The simulations confirmed a fluctuation in feed-end temperature

Fig. 5 - Effect of lime-mud moisture content variations on lime-kiln feed-end temperature during stable fuel feeding (100% heating oil at 32.5 MW) and at production rate (415 tons/day).

Fig. 6 - Effect of a sudden production increase on 2016-11-19 under stable production conditions and fuel supply (100% heating oil). The new calculation refers to the production rate as described in Eq. (5). FET denotes the feed-end temperature of the lime kiln. 6b is a magnification of 6a.


of around 3°C–5°C corresponding to fluctuations in moisture content, as shown in Fig. 5. Furthermore, the simulation also showed that even though the moisture content did indeed induce fluctuations in feed-end temperature, these fluctuations did not propagate and did not pose any problem for recarbonation in the lime kiln. The simulation of the sudden pro-duction increase shown in Fig. 6 indicates a decrease in feed-end temperature of around 5°C, which is a considerably lower decrease than was seen in the mill data, where the temperature dropped from 700°C to 670°C. However, the simulation assumed that all lime mud was carried with the hot flue gas to the flash dryer, but in reality part of the moist lime mud falls di-rectly into the kiln feed end. This is in fact an interesting simulation application that could be used to estimate the amount of mud that falls through into the kiln feed. However, such an experiment is beyond the scope of this study and was not inves-tigated further.

RESULTS AND DISCUSSION OF MILL TRIALS

To investigate the actions and conse-quences associated with the observed fluc-

tuations in lime-mud feed rate and mois-ture content induced in a lime-mud filter, two mill trials were conducted using NIR measurements and a band scale.

Trial 1During the first trial, the water flow to the tray, which was used to maintain a steady level, was shut off (Fig. 7). The purpose of the trial was to investigate how the wa-ter affected the lime-mud moisture con-tent and feed rate after the filter. The level control was believed to cause a variation in slurry dilution and an uneven distribution of both lime-mud dry solids content and density in the tray.

However, the results show that the standard deviation in moisture content ac-tually increased from 0.5 (average of 26.3) to 0.9 (average 26.2) percentage points when the level in the filter was not con-trolled. An interesting observation is that the moisture content was not significantly affected even though the vacuum went from -60 kPa to -54 kPa. This apparently contradictory result can be explained by the fact that two mechanisms that influ-ence moisture content were affected at the same time. First, the drop in vacuum de-creased the driving force for dewatering. Second, the residence time on the filter

cloth increased as the level in the tray de-creased, which enhanced dewatering. The reason for the weaker vacuum was prob-ably that a larger surface of the drum was exposed to air as the liquid level became lower. The greater access to air increased pressure equalization in the drum. Never-theless, not diluting the lime-mud slurry before its uptake on the filter appears to have had a positive effect on the dry lime-mud feed rate to the kiln. This had a positive effect on feed-end temperature stability because the influx of lime mud varied less. The water flow for level con-trol was measured as between 12.5 and 14.5 litres/s. The conclusion of the trial was that the level control does not appear to be necessary and could be shut off to stabilize the dry-mud feed rate and also make it easier to maintain high green li-quor strength while also saving water.

Trial 2As shown in Fig. 6, the washing of the doctor blade induces instability in the dry lime-mud feed rate. Therefore, in the sec-ond trial, the washing sequence was shut off. This was achieved by covering the back side (as well as the top side) in a non-stick material. When the washing was shut off, the induced spikes disappeared.

CONCLUSIONS

A new procedure for calculating the pro-duction of available CaO was developed. This calculation requires the mass flow, moisture content, and carbon content of the lime mud after the lime-mud filter together with the carbon content of the reburned lime. During implementation of this new calculation method, previ-ously unnoticed process variabilities were discovered. Unexpected fluctuations in lime-mud feed rate and moisture content were observed. These fluctuations were revealed as minor feed-end temperature variations in the lime kiln. The tempera-ture variations induced were generally too small to pose any major problems for maintaining satisfactory stability in a kiln equipped with a flash dryer. Further-more, when using the new procedure for

Fig. 7 - Dynamics of trial 1. The fuel supply and lime-mud feed rate after the filter were constant (32.3 MW and 38.8 tons/hour). The vacuum went from -60 kPa to -54 kPa. The distinct spikes in the dry lime-mud feed rate were caused by the doctor-blade washing sequence.



calculating reburned lime production, short periods with a large production in-crease were noticed when the doctor blade was washed, which led to lime mud falling directly into the kiln instead of being car-ried to the flash dryer. This introduction of moist mud into the kiln caused a distinct pulsation in feed-end temperature. Doc-tor-blade washing was then shut down to promote kiln stability. This was achieved by dressing the doctor blade in a non-stick coating. The lime-mud filter level control could also be shut off. These measures were beneficial for saving water and pro-moting stability in green liquor TTA and dry lime-mud feed rate and subsequently in lime-kiln feed-end temperature.

The proposed calculation procedure also proved to be a practical tool for pre-dicting available CaO, which is helpful to avoid overliming.

ACKNOWLEDGEMENTS

Financial support received from Södra Skogsägarnas Stiftelse för Forskning, Ut-veckling, och Utbildning is gratefully ac-knowledged. Special thanks are given for the valuable work, patience, and under-standing of the staff at Södra Cell Mön-sterås Mill.

Rydin, S., Haglund, P., and Mattson, E., “Causticizing of Technical Green Liquors with Various Lime Qualities,” Svensk Papperstidning, 80(2): 54-58 (1977).Hanson, C. and Theliander, H., “Prop-erties and Quality of Lime, Part 1. The Influence of Conditions during Reburn-ing”, Nordic Pulp and Paper Research Journal, 9(3):161-166 (1994).Tran, H., Mao, X., and Barham, D., “Mechanisms of Ring Formation in Lime Kilns”, Journal of Pulp and Paper Science, 19(4): J167-J174 (1993).Eriksson, G. and Theliander, H., “Dis-placement Washing of Lime Mud”, Nor-dic Pulp and Paper Research Journal,

1.

2.

3.

4.

REFERENCES

9(1): 60-66 (1994).Lindblom, J., Wildt, J., and Theliander, H., “Sintering of Calcined Lime Mud:The Influence of Sodium and Sulphate Content in the Lime”, Nordic Pulp and Paper Research Journal, 19(11): 23-30 (1998).Wallbäcks, L., Theliander, H., Johanzon, O., Pettersson, T., and Welin, A., “Stable Lime Kiln Feed through On-Line Mea-surements and Feed Control”, Proceed-ings, International Chemical Recovery Conference, Session 2B, 170-186 (2014).Theliander, H., “On the Equilibrium of the Causticizing Reaction”, Nordic Pulp and Paper Research Journal, 7(2): 81-87 (1992).

5.

6.

7.

VARIABILITY IN LIME-MUD FEED RATE INDUCED IN THE LIME … · in the lime kiln [3]. Furthermore,...

Documents

Transcript of VARIABILITY IN LIME-MUD FEED RATE INDUCED IN THE LIME … · in the lime kiln [3]. Furthermore,...